The body of the robot comprises two end frames and an intermediate stiffening frame linked by four longitudinal spars. Each leg is powered by two RC servo motors. Swing motors move the legs backwards and forwards. Lift motors lift the motors up and down. The swing motors are rigidly mounted inside the chassis. The lift motors are fitted inside a folded sheet metal frame which is suspended from the swing motor output shaft. With the benefit of hindsight, not a very satisfactory arrangement.
The frames are cut from 0.6mm aluminium sheet which is then folded in a vice to give stiffening flanges. The frames were lightened by punching 32.5mm holes in the centre with a sheet metal punch. This size just happened to be the largest stocked by my supplier.
The swing motors were mounted inside the chassis on cross members made from 0.6mm sheet folded to form a stiffening flange and epoxied to the longitudinal tubes.
The lift motor frames were made from folded 0.6mm aluminium sheet, with numerous holes punched in it to reduce weight. A cut down servo horn was attached to each frame with hot melt glue and a small self tapping screw. This horn was used to suspend the frame from the swing motor output shaft. The lift motors were held in the frame with a couple of small cable ties.
The battery pack is fixed with velcro pads to on a mounting plate fitted between the motor bearing members.
The longitudinal spars are cut from 5/32" OD aluminium alloy thin walled tubes. These tubes are manufactured by K & S Engineering of Chicago and can be obtained from most modelling shops.
The simple "rowboat style" legs are made from 90mm lengths of 5/32" OD aluminium tubes. A flange made of 20 x 10 x 2mm aluminium bar is epoxied to one end of the tube and is shaped to fit into the socket of a sliding arm servo horn.
I followed the tip from Mobile Robots[3] on setting out sheet metal components. I drew the developed view full size using a drawing program ( CorelDraw) on a PC, printed it on a laser printer and then gummed it to the aluminium sheet. This is then used as a template for drilling, cutting out and folding the component. Some people have found that diagrams printed this way on a laser printer are slighly distorted. This effect is mainly due to the paper stretching as the toner is baked on it and can be mitigated by first printing a blank sheet and then reusing this sheet to print the diagram. In any case the dimensions of these metal components are not sufficiently critical for this to matter.
| Frames 12K | Lift Support Frame 31K | Spar Leg 8K |
The body of the robot comprises two end frames and an intermediate stiffening frame linked by four longitudinal spars. The leg assemblages are mounted between top and bottom traverse bearers which slide along the spars.
Each leg assemblage comprises two orthogonally mounted RC servo motors. One, the swing motor is used to swing the legs backwards and forwards; the other, the lift servo is used to lift the foot up and down. A four bar linkage is used to constrain the vertical foot motion to a near straight line. The output shaft from the swing motor is fixed to the upper bearing mounting so that when the motor turns, the body of the motor rotates with respect to the robot chassis. The bottom casing of the swing motor is attached to the bottom bearing assembly by a spigot bearing to minimise any transverse bending moments on the motor drive train.
The frames are fabricated from 0.6mm sheet aluminium. Most of the interior area is removed to reduce weight and lips are folded to form stiffening flanges.
The longitudinal spars are cut from 7/32" OD aluminium alloy thin walled tubes. These tubes are manufactured by K & S Engineering of Chicago and can be obtained from most modelling shops.
The traverse bearers are made from 8.2 x 9.2 x 0.85mm extruded aluminium U section.
The swing motors are fitted with a sliding arm mounting horn which is attached to the upper bearing assembly with self tapping screws.
The lower bearing assemblies for the swing servo motors are the most complicated mechanical parts. The upper part of the bearing is an aluminium pad and spigot which is epoxied to the bottom of the servo casing The spigot seats in an aluminium socket which is epoxied to the bottom bearing frame. I turned the aluminium pad and spigot and socket from 10mm square bar.
The swing motors are attached in pairs to acrylic yoke bars by sliding arm servo horns. The yoke bars are slung from the upper bearing frames with a central nylon bolt and an aluminium shear pin to prevent the yoke rotating with respect to the chassis.
The swing and lifting servos are joined with double sided sticky tape reinforced with a cable tie around the pair.
The leg members are made from 5/32" OD drawn thin walled aluminium alloy tubing. This can be obtained in foot lengths from aircraft model shops. A clevis and pin joint is used to allow sufficient angular movement for the linkage joints. The clevises are cut from 6mm acrylic sheet. The leg members are reinforced with a transverse sleeve made from 3/32" OD aluminium tubing glued in place with epoxy to take the 1/16" OD tubing used for the clevis pins. The lower driving link is made of 5/32" tubing epoxied to a 20mm length of 8.2 x 9.2 x 0.85mm U section to which a cut down servo horn is attached with self tapping screws.
The upper fixed point for the leg four bar linkage is made from extruded aluminium 8.2 x 9.2 x 0.85mm U channel section. This is fixed to the lift servo mounting points using M2 nuts and bolts and acrylic spacers.
The problems with this layout were:
| Frames 26K |
The body of the robot comprises two end frames linked by four longitudinal spars. The leg assemblages are mounted between top and bottom traverse bearers which slide along the spars.
Each leg assemblage comprises two orthogonally mounted RC servo motors. One, the swing motor is used to swing the legs backwards and forwards; the other, the lift servo is used to lift the foot up and down. A four bar linkage is used to constrain the vertical foot motion to a near straight line. The output shaft from the swing motor is fixed to the upper bearing mounting so that when the motor turns, the body of the motor rotates with respect to the robot chassis. The bottom casing of the swing motor is attached to the bottom bearing assembly by an adjustable bearing to minimise any transverse bending moments on the motor drive train.
The end frames are fabricated from extruded aluminium sections with are glued together with epoxy. The sections were obtained from the local DIY superstore.
The longitudinal spars are cut from carbon fibre composite tubes obtained from the local kite shop. Very stiff and light thin walled spiral wound tube ( Carbon composite arrow shafts can be used but although they are much cheaper they are a lot heavier and have a lower bending stiffness. Since they are pultruded they also have poor torsional strength and stiffness). The use of these tubes produce a sufficiently stiff structure that the intermediate stiffening frame used in earlier designs can be discarded. This reduces weight and allows more flexibility in the positioning of the motor and battery assemblies. The longitudinal members are fastened to the end frames with set screws into acrylic blocks which are epoxied to the end frames. Short lengths of aluminium alloy tube are epoxied into the ends of the carbon fibre tubes to locally reinforce them.
The traverse bearers are cut from 6mm thick acrylic sheet. This is one of the names for polymethylmethacrylate (PMMA) and sold under various trade names such as Perspex, Plexiglas, Oroglas and Lucite etc.
The end frame mounting blocks and the transverse bearers are held in place with M2 set screws. Small pads of thin brass shim are used to protect the soft epoxy of the carbon tubes from local damage by the screw ends. The set screws could be omitted and the transverse bearers and mounting blocks could be epoxied to the spars, but this arrangement permits adjustment and allows for the components to be reused which is not an inconsiderable factor when the carbon fibre tubes cost £15 per metre.
The swing motors are fitted with a sliding arm mounting horn which is attached to the upper bearing assembly with self tapping screws.
The lower bearing assemblies for the swing servo motors are the most complicated mechanical parts. The upper part of the bearing is an aluminium pad and spigot which is attached to the bottom of the servo casing with a foam sticky pad. The sticky pad provides some compliance and is easily removed if the motors need to be recycled for something else. The turned spigot on the bottom of the pad seats in a nylon bush. This bush is formed from a section of 10mm OD nylon rod which has a 4.8mm diameter axial hole bored through it and has an external M10 thread. The bushing screws into a lower bearing pad of 6mm acrylic which is glued onto the end of the lower transverse bearer. This arrangement provides a low friction bearing along with a vertical adjustment mechanism. I turned the aluminium pad and spigot from 10mm square bar, but it could have been fabricated from 2mm acrylic sheet and a 12mm length of 3/16" OD aluminium alloy tube if I hadn't had access to a lathe.
The swing and lifting servos are joined with double sided sticky tape reinforced with a cable tie around the pair.
The leg members are made from drawn thin walled aluminium alloy tubing. This can be obtained in foot lengths from aircraft model shops. Two diameters 5/32" and 3/16" are used. A clevis and pin joint is used to allow sufficient angular movement for the linkage joints. The clevises are cut from 6mm acrylic sheet. The leg members are reinforced with a transverse sleeve made from 1/8" OD aluminium tubing glued in place with epoxy to take the 3/32" OD tubing used for the clevis pins. The lower driving link is made of 5/32" OD tubing epoxied to a 20mm length of cut down 8.2 x 9.2 x 0.85mm U section to which a cut down servo horn is attached with self tapping screws.
The upper fixed point for the leg four bar linkage is made from extruded aluminium 8.2 x 9.2 x 0.85mm U channel section epoxied to an acrylic spacer which is attached to the lift servo mounting points with self tapping screws.
The foot switches are stuck to the ends of the legs with RTV silicone which provides a strong yet slightly compliant resilient joint.
The electronics PCBs and battery pack are clipped on to the longitudinal spars with plastic mains cable clips with the nails removed and the clips fixed to the PCB/battery pack with hot melt glue.
Numerous holes are drilled into anything that looked solid. I say I did it to reduce weight, but since the total weight reduction was only about 15g, perhaps I did it because it looks better!. And it does provide plenty of positions to bolt sensors on.
The acrylic clevis joints are fiddly to make and are not strong enough. They often broke and had to be reinforced with small strips of acrylic which were cemented on to the lower end where the aluminium tube entered the joint.
| End frame 11K | Side elevation 5K | Top elevation 5K |
| Front view of leg assembly 12K | Side view of leg assembly 10K | Exploded view of Lower Bearing 5K |
| Leg Joint 8K |
The upper and lower swing motor bearers follow the same design as in Rodney 3 except that the longitudinal carbon composite tubes are held in place with pinch bolts instead of set screws. The upper and lower bearers are now joined with a section of acrylic so that the new combined frame looks like an H on it's side. The aluminium section end frames are discarded and the longitudinal spars reduced in length by 80mm since the front and back motors are now mounted at the ends of the tubes.
It's quite difficult to get hold of thickish acrylic sheet in small quantities. A cheaper alternative might be to replace the acrylic H frames by ones cut from 5mm 5-ply birch plywood. The longerons would then be glued in place rather than clamped and care would be needed in screwing the nylon sliding horn pieces to the weak edge of the plywood. ( You might have to drill out and fill oversize holes which could then be drilled for the self taps). Plywood is not a good material for tapping the M10 thread for the adjuster bushes so the lower bearing pads would still have to be made of acrylic or some other suitable material and then glued into place on the plywood frame. Cut out the three frames with a jig or fretsaw, clamp them together and then gang drill all the tube holes to ensure correct alignment.
The cost of the chassis can be very much reduced by replacing the 7mm spiral wound carbon composite longerons by ones made of either 9/32" OD aluminium tubes or carbon composite arrow shafts.
I have now come across kite spars made from a mixture of glass and carbon fibres which are very light and stiff and cheap at only £3.75 for a length of 825mm 9mm OD tube. A smaller diameter spar is available for £3.50 which is an even better bet. At this price I would make the length of the longerons 205mm so that they can be cut from one spar and then permamently glue the acrylic or plywood H frames in place with expoxy, dispensing with the pinch bolt clamps. 205mm is a bit short; 240mm would be better, so the Rolls Royce solution would be to cut the lengths from two spars.
Although the use of aircraft modeller's steel clevises makes a neat looking leg with no appreciable slop, the aluminium clevis inserts need to be turned and milled and threaded. Each leg has five glued joints and it is difficult to make good glued joints in aluminium without extensive preparation. Bearing in mind that six legs are required for each robot, this design fails the "Keep It Simple" adage and is too much like hard work!.
If the upper and lower tubular links are replaced by 8.2 x 9.2 x 0.85 mm aluminium U section, a 1/4" OD tube with thin nylon washers will fit snugly into slots cut in the end of the section for the main leg spar. The length of the main leg spar is such that two members can be cut from a one foot length of tubing. The weight of the U section members can be reduced by drilling numerous lightening holes. The transverse sleeves and pins for the leg spar are made from 5/32" and 1/8" OD K & S brass tubing since the aluminium tubing used in the Rodney 3 design wore rather quickly. The fitted pin needs to be held in place in the U section with a blob of cement to prevent fretting between the pin and the soft aluminium enlarging the holes in the U section. The acrylic spacer and U section inboard support for the upper link member is replaced by a longer acrylic spacer with a cantilevered bearing made from a 26mm length of 1/8" OD brass tubing. The upper link is held in place on the brass tube with nylon washers and a brass model aircraft wheel collar. With this design the machining for each leg is reduced to cutting the members to length, drilling twelve holes (plus lightening holes to taste) and cutting slots in the outboard ends of the links. As well as being easier to make, the larger diameter tubing and U channel make for a stronger more robust leg unit. Each leg assembly with lightening holes weighs about 12 g.
The most difficult part is now the threading operations for the lower bearing assemblies. These can be avoided by using the brass bushings made to support long 1/4" spindles on instrument panels. The upper part of the bearing is made from a 14mm square pad of 2mm thick acrylic sheeting centrally drilled to take a 14mm length of 1/4" OD K & S aluminium tubing for the bearing spigot. The bush retaining nut is expoxied to a drilled block of acrylic ( or plywood) for attaching to the H frames. Since the retaining nut is quite a loose fit on the bush, the bush threads can be wrapped with a length of plumber's PTFE tape to take up the play. The only drawback to this design is that, at about 6g each, it is a lot heavier than the original.
The main effect of the design changes between Rodney 3 and Rodney 4 with the Mk 2 leg design is to greatly simplify the construction without significantly decreasing the robot's robustness or increasing it's weight, and at the same time increase the stiffness and robustness of the leg units.
All the parts for this design can be cut to size using a junior hacksaw. A M10 tap and die are needed for the lower bearing assembly original design. However the alignment and positioning of the drilled holes is critical to the design and this is best achieved using some form of drill stand.
The current Rodney perhaps should be more correctly called 3.75 in that it uses the body of Rodney 3 with the Mk2 improved leg design from Rodney 4.
| New Frame 6K | Plywood Frame 3K | Mk 1 Improved Leg 7K | Mk 2 Improved Leg 6K | Exploded view of Alternative Lower Bearing 5K |
Rodney 1 was built using DIY simple hand tools; an electric drill, an electric jigsaw and a junior hacksaw.
I cut the aluminium sheet for Rodney 1 using a hand electric jigsaw fitted with a fine metal cutting blade. I gummed the developed view of the part onto the sheet metal, clamped the metal sheet to a thin backing board of hardboard to stiffen it and then cut the shape out.
After Rodney 1, I borrowed a small bandsaw ( thanks Mike!) which made it all a lot quicker cutting the extruded sections and fabricating the parts cut from acrylic sheet. A fine toothed handsaw (preferably a panel or even a tenon) could be used instead of the jig saw to rough cut the acrylic. A bandsaw is a very useful tool for fabricating the parts cut from acrylic sheet. A hand fretsaw was used to remove the cutouts from the end frames for Rodney 2.
The aluminium and carbon composite tubes were cut to length using a junior hacksaw with a mandrel of piano wire inside them to prevent ovalising and crushing the tube in the vice.
Most of the construction of this project involved drilling lots of holes. For Rodney 1 which was constructed out of thin aluminium sheet and aluminium tubes, I used an electric hand drill to drill all the holes and it worked fine. But when you start using thicker materials like the acrylic sheet and the extruded aluminium sections, it's a lot harder to keep the hole axis perpendicular to the surface of the sheet without a drill stand and if you're going to thread a stiff carbon fibre spar through five close tolerance holes over a distance of 320mm, the holes have to be drilled perpendicular to the surface and in the correct position. But the most critical operation is cross drilling the holes for the bearing sleeves in the legs, particularly for the tubular members. For the finished leg to behave correctly, the following conditions must be satisfied:
This is very difficult without some sort of drill stand, so I bought a pillar drill. I started by making a simple jig to clamp the tubes whilst drilling them. I later bought a compound drill vice which speeds up the drilling of multiple holes in a component no end and is very useful to position the bearing sleeves holes. A pillar drill is also very useful when tapping holes for the set screws.
In the plywood H frame design for Rodney 4, the bottom limb of the frame denies access to the top limb to drill holes for the self tapping screws used to fix the sliding servo horn. These holes can be drilled with a pin chuck or failing that a drill bit and a pair of pliers would suffice to drill into wood!.
For Rodney 3, the end frames fixing blocks and the motor bearing members are clamped to the longitudinal tubes with M2 set screws. A M2 tap was used to cut the threads in the acrylic. A pillar drill is an invaluable alignment guide when tapping holes!. A M10 tap was used to cut the thread in the lower bearing pad and the nylon adjustment bush was threaded using a M10 die.
In Rodney 4, the set screws are replaced by a pinch bolt arrangement, so the M2 tap is no longer needed. The M10 tap and die is not needed if preformed brass bushes are used for the lower brearing assembly.
A lathe was used to turn the spigot and part the bearing pad from 10mm square bar for the lower bearing pad and spigot. It was also used to bore and thread the nylon adjuster, but that could have been done with a lot of care using the pillar drill. To see if the lathe work could be avoided, I made a perfectly good pad and spigot from a short length of shop bought small diameter (11/64") acrylic rod glued into a small pad of thin (2mm) acrylic cut from sheet. I have since found that 3/16" aluminium alloy tubing is a better material for the spigot since it is more perfectly round than the acrylic rod.
In the Mk 1 leg design for Rodney 4 a lathe was used to turn the 4mm diameter aluminium rod to fit in the 3/16" OD aluminium tubes, thread the insert for the M3 clevises and to mill flats on the rod to accomodate the clevises. No turning operations are needed for the Mk2 leg design.
